基于TSVM的燃机SFC整流晶闸管故障诊断
TSVM Based Fault Diagnosis for SFC Rectifying Thyristors in Combustion Gas Turbines
侯苏宁 1陈俊儒 2张丹青 1陈露垚 1史华仁 1奚新国1
作者信息
- 1. 华能南京燃机发电有限公司,江苏 南京 210046
- 2. 南京南瑞继保电气有限公司,江苏 南京 211102
- 折叠
摘要
将燃机SFC整流晶闸管故障诊断抽象为多分类问题,提出了基于TSVM的分层多分类策略用于故障诊断、故障类型判别及故障定位.在燃机SFC整流桥机理研究及仿真分析基础上,归纳了故障诊断模型、故障类型判别模型及故障定位模型的输入参数.构建了SFC整流电路仿真模型,形成了 22 个案例,并调整触发角进行了共 2200 次仿真,获取了数据集用于模型检验.结果表明,所提出的策略在故障诊断、故障类型判别及故障晶闸管定位方面都取得了较高准确率,其可行性和有效性得到了充分验证.
Abstract
In this paper the fault diagnosis of SFC rectifying thyristors in combustion gas turbines are abstracted as mul-ti-classification tasks.A TSVM based multi-classification strategy is then proposed for fault diagnosis,fault type identification and faulty thyristor location in a hierarchical framework.With deep research on SFC rectifier bridge and simulation analysis,the input vectors for fault diagnosis model,fault type identification model,and fault location model are elaborately construct-ed.A simulation model of SFC rectifier bridge is constructed.Through fire angle regulation,22 cases are formed,and 2200 simulations are conducted.The obtained data sets are then used for model verification.The results show that the proposed strategy gets considerably high accuracies in fault diagnosis,fault type identification and fault location,respectively.
关键词
静止变频器/晶闸管故障诊断/孪生支持向量机Key words
static frequency converter/fault diagnosis for rectifying thyristors/twin support vector machine引用本文复制引用
基金项目
中国华能集团科技项目(HNKJ21-HF265)
出版年
2024
NETL
NSTL
万方数据